Introduction
Trials of brief interventions on alcohol have predominantly recruited participants through health-care settings, particularly primary care.8,9,36 As this study was aimed at community-dwelling individuals, it seemed sensible to use the electronic patient records systems held by GPs. However, this method may not be sufficient to recruit the target group, men aged 25–44 years. These men are usually healthy and are seldom in contact with health services, so an alternative recruitment method is needed.
This chapter reviews the literature on techniques to increase participation in research studies to identify strategies to improve recruitment. It then describes a recruitment strategy used in parallel with recruitment through primary care, and reports on strengths and weaknesses of each method. In view of the anticipated difficulty of recruiting disadvantaged young to middle-aged men, RDS was used alongside recruitment through primary care. RDS provides a method for recruiting hard-to-reach groups.37,38 The technique assumes that the target population is distributed through a number of socially networked groups and is thus suitable for a group behaviour such as drinking.
Techniques to promote recruitment
Several techniques have been shown to increase recruitment rates. As non-contact makes at least as large a contribution to failure to recruit as does refusal to participate,105,106 both non-contact and refusal were addressed. Systematic reviews show that repeated attempts at contact and monetary incentives increase recruitment to research studies.45,107 Strategies which have been found to be effective in general population groups31,32,44,107–110 are shown in Box 2. Similar strategies have been identified for recruiting hard-to-reach and minority groups, although particular emphasis is given to culturally sensitive study materials, multiple recruitment strategies and incentives.111–115
Effective techniques to increase response rates across all studies Financial or other appropriate incentives
Methods
Participants
Men aged 25–44 years living in areas of high deprivation were recruited. Deprivation was measured using the SIMD,46 which is similar to the English Index of Multiple Deprivation.116
Inclusion/exclusion criteria
Men were included in the study if they had two or more episodes of heavy drinking (≥ 8 units in a single session) in the preceding month. Men currently attending care at an alcohol problem service and men who would not be contactable by mobile telephone for any part of the intervention and follow-up period were excluded.
Recruitment methods
The invitation letter, participant information leaflet and consent form are presented in Appendix 2. The screening questionnaire is in Appendix 3.
Strategy 1: recruitment through primary care
Potential participants were identified from the practice lists of three GPs. These lists provide data on age, telephone numbers and postcode, which facilitate selection of men aged 25–44 years from areas of deprivation. Postcode was used to derive the SIMD score.117 Potential participants randomly selected from the two most disadvantaged deciles were sent a letter by their GP inviting them to take part. The letter was personally addressed, mentioned the University of Dundee and stated that a financial reward would be given. The accompanying participant information leaflet carried the university's logo and stressed the confidentiality of the study. An opt-out strategy was used for recruitment. The names, addresses and telephone numbers of those who did not decline to take part were provided to the researchers by the GPs. Contact with these individuals was made by telephone approximately 2 weeks after the GP letter. The researcher described the study to these individuals, answered any questions and asked them if they wished to participate.
Up to six attempts were made to contact by telephone the men who did not opt out, at different times of the day and on different days of the week. Those who agreed to take part when contacted were asked to complete the screening questionnaire to determine eligibility. Ethical consent was then obtained from suitable individuals. On completion of the baseline questionnaire, participants were sent an initial £10 gift voucher to offset any charges incurred by receiving and responding to text messages. They were also sent a £5 gift voucher for each week of the study and a £10 voucher for completing the outcome assessment.
Strategy 2: respondent-driven sampling
Respondent-driven sampling begins with the identification of ‘seed’ individuals obtained from different locations. In this study, seeds were recruited from several venues in areas of high deprivation. The seed individuals identified suitable subjects from their social networks and recruited them to the study. A key element of the technique was the use of incentives to each seed person for taking part in the study and for each of the individuals they recruited. All participants recruited by this method also received the sequence of gift vouchers as described in recruitment strategy 1. In addition, the seed individuals received a £5 gift voucher for each person they recruited.
Potential seed individuals contacted directly were told about the nature of the study, the financial incentives and the sponsorship by the university. They were also given participant information leaflets as described for recruitment strategy 1. Individuals nominated by seeds were given the same information by telephone and were sent a copy of the participant information leaflet by post. This was followed up by another telephone call to explain the nature of the study and obtain informed consent as described below.
Initial screening and informed consent
Individuals identified by the recruitment strategies were screened by a telephone call from a researcher to establish current drinking levels. Those who reported binge drinking (≥ 8 units in a single session) at least twice in the previous 30 days were identified as potential participants. The details on the participant information leaflet were explained, such as their right to withdraw from the study at any time and the confidentiality of the data. Verbal consent was sought after potential participants confirmed that they had understood the contents of the information leaflet. Those who agreed to participate were sent a text message which asked them to reply to the message if they wanted to take part in the study. This meant that potential participants had to indicate their consent to participate by taking the positive action of responding to a text message. During the screening interview a consent form was completed by the research fellow. When the consent text was received, the date and time of obtaining consent was recorded and the form was signed by the research fellow. The consenting text was retained as proof of consent.
Results
Recruitment
Recruitment began on 17 March 2011 and was completed on 12 June 2011. A total of 67 men were recruited, exceeding the target of 60. The over-recruitment occurred because of the yield from the final GP. As the response rate could not be predicted, letters of invitation were sent to all 120 men randomly selected from the GP list at that practice. Recruitment could have been stopped when the intended target of 60 participants was achieved, but it seemed unwise, and possibly unethical, to write to men about a study then subsequently fail to invite them to participate.
Yield from recruitment through primary care
Identifying men living in areas of deprivation from practice databases proved straightforward. A random sample was taken of men on the practice lists who were in the appropriate age and deprivation categories. The sample taken allowed for anticipated loss due to failure to contact, ineligibility (do not binge drink frequently) and refusal. GPs screened these samples to exclude candidates because of concerns about health or family problems. The GPs noted that screening the lists of men was difficult because few of the men attended their doctor regularly. Almost all the men (89%) had telephone numbers, either landline or mobile, in their GP records and few (3%) were excluded by their GP ().
Initial recruitment details
As expected, some of the telephone numbers were invalid (); in fact, 25% of numbers tested were invalid. As up to six attempts were made to contact the men (covering daytime, evenings and weekends) it was expected that most men would answer the telephone. In the event, 18% of those with valid telephone numbers did not answer the telephone. Of those contacted, 31% refused to participate and a further 40% were found to be ineligible because they did not binge drink regularly. As a result, 29% of those contacted were recruited.
Results of attempts at telephone contact
Lessons from recruitment through primary care
Recruitment through primary care is subject to losses from failure to contact (missing or incorrect telephone numbers) or refusal to take part. Ineligibility, because of low frequency of binge drinking, was a major reason for failure to recruit. Some men appeared concerned that the approach might be part of a scam to steal money. This was more marked when first-approach telephone calls came from an institutional 0845 telephone number. Use of a dedicated study mobile telephone number overcame this difficulty. As participants are randomised to treatment group, losses to recruitment will not affect internal validity of the study. Trials recruiting through this route should contact substantially more individuals than required to allow for these losses.
Yield from recruitment through respondent-driven sampling
In total, 30 men were recruited over a 5-week period. The sample of 30 men comprised 12 who did not nominate any friends and seven who nominated 11 friends in total. All of the nominated friends were recruited to the study. The venues through which men were recruited are shown in (nominees have been allocated to the venue on the nominator). Multiple locations were used to ensure recruitment of individuals from different social networks. Recruitment took place during the day, in the evening and at the weekend.
Venues and recruitment for RDS
Lessons from the respondent-driven sampling
Modes of recruitment
In total, 30 men were recruited from a variety of venues. More men were recruited from some venues than others, but interpreting this is problematic. At venues where groups of men were present it was not possible to count the number approached because they entered and left together. For example, teams of men at sports centres would be leaving at the same time. Thus, it was not possible to identify who had heard the initial request for attention, who had to leave quickly because of other engagements and who was not interested in the study.
Young and middle-aged men are willing to be stopped opportunistically at a wide range of settings. Approaching small groups was more productive than approaching single men, possibly because of the security of belonging to a group. In a group, the first response made, either in support of or against the study, often set the tone for the whole group. Timing is crucial for recruitment through sporting activities, as the men often wish to hurry away when the activity is finished.
Recruitment at some settings (e.g. snooker halls and slot machine venues) proved problematic. Staff at these venues wished to protect their clients from being disturbed. This was quickly established at the first contact and these venues were removed from the list. At other settings, such as community centres and sports clubs, many or all of the men were ineligible because they did not binge drink frequently.
Further education staff can greatly expedite the recruitment of men through their settings; they know and are known by the potential recruits, giving an opportunity for a personal touch in recruitment. Recruitment through work settings can be effective. However, in contrast to the community and further education settings, involvement of managerial staff in workplace settings could be counter-productive. Although the managers were keen to help identify men, recruitment was poor. It is possible that the men were concerned that information on their alcohol consumption would be passed to their employers. An alternative strategy, where the manager gave approval but the researcher did the recruitment, proved more successful.
Some bar staff were willing to nominate regular customers, although many other staff were reluctant to do so. Although there were few men in shopping centres, informal approaches were potentially useful, as the men appeared to have more time to chat. There were very few single men or groups of men in cafes.
Nomination of peers
Each of the recruited men was asked to nominate friends for possible inclusion in the study. Despite being offered financial reward for this (£5 for each friend nominated), only seven men did so. The rate of nomination was lower than in previous reports.118,119 It is possible that the focus of the study on binge drinking inhibited men from nominating friends. Alternatively it could be that the nominators did not know enough about the study to feel confident in nominating a friend. Interestingly, one man who nominated two friends was never recruited to the study although he had said that he would take part.
Distribution by deprivation category
Deprivation was measured by the SIMD, in which decile 1 is high deprivation (). Although most of the men lived in areas of high deprivation, a few were from areas of low deprivation. Assessing the socioeconomic status of potential recruits during the recruitment proved problematic. It was decided it would have been impolite to ask the men directly whether or not they were from disadvantaged areas. In a full efficacy trial it would be possible to conduct a subgroup analysis which excluded the men from the least deprived areas. This could be accompanied by over-recruitment to ensure the intended sample size is achieved after the exclusions.
Yield from RDS by deprivation category
Additional venues for recruitment
Many people in the target group (binge drinking young and middle-aged socially disadvantaged men) are hidden because they do not participate in community activities. It proved difficult to recruit these men from community-based activities or sports centres, even though they are located in areas of high deprivation. Thus, instead of recruiting from organised activities, effort should be directed to venues in areas of high deprivation where these people must go, such as:
Other lessons learned
Several observations were made during the conduct of the telephone calls. As these were unexpected findings, data recording procedures had not been put in place to record them. Thus, they could not be quantified.
Some men found the information given about the study did not fully explain what would be involved. Although further information was given during recruitment, it is possible that some men may not have asked for clarification, and instead refused to participate. Forming a rapport at the first telephone call and actively exploring concerns about participation can prevent this loss of participants.
People who drink a lot may be unwilling to disclose this to a stranger. They would thus be classed as not meeting the entry criteria. Establishing rapport, emphasising the confidential nature of the data and stressing the benefits of the research will minimise this problem.
Discussion
This study has shown it is possible to recruit sufficient disadvantaged men in early mid-life using recruitment through GPs or through RDS. Although this study focused on a difficult-to-recruit group it easily exceeded its recruitment target. Based on previous research,18 it is likely that the use of incentives contributed to the successful recruitment. The other recruitment techniques, particularly personalised approaches, credibility of the source and assurances of confidentiality, are likely to have played their part.
Both methods of recruitment proved successful. Recruitment through primary care is a well-tested strategy for trials of brief interventions for alcohol. In this study the intended sample size was easily attained, although only a subsample of men from the three GPs were approached. It also showed that failure to contact was a greater cause of non-inclusion than was refusal to participate. Recruitment through RDS was also successful in recruiting disadvantaged young men. This method has the potential to recruit large numbers of individuals, because of the many venues from which participants could be recruited. This method of recruitment requires a period of preliminary fieldwork to identify suitable venues and to negotiate approval to recruit through them.
This study was designed with a view to national roll-out. To keep costs low, a national study would require a low level of staff involvement in recruitment. Thus, the study developed and tested a non-contact method for recruiting and obtaining informed consent. This method proved successful. An additional benefit of the minimal-contact recruitment strategy is that the process of entry to the study does not become part of the behaviour change intervention. Thus, the intervention tested in this feasibility study and the proposed full trial will be the same as that used in any national roll-out.
